The present paper describes a new approach to the problem of compensation for language modeling errors adapted for key-spotting systems using infill units. A regular approach to compensation for language modeling errors doesn't take into account the difference in phonetic length of key-words and infill units. The suggested approach is free of this disadvantage and we believe that it will make possible to decrease rather a high level of false alarms, which is quite typical for key-spotting systems with infill units.
Cite as: Zulkarneev, M.Y. (2004) An approach to compensation for language modeling errors in the key-spotting systems. Proc. 9th Conference on Speech and Computer (SPECOM 2004), 348-350
@inproceedings{zulkarneev04_specom,
author={Mikhail Yu. Zulkarneev},
title={{An approach to compensation for language modeling errors in the key-spotting systems}},
year=2004,
booktitle={Proc. 9th Conference on Speech and Computer (SPECOM 2004)},
pages={348--350}
}